CANADIAN JOURNAL OF BEHAVIOURAL SCIENCE, 1999, 31:3, 150-160
GENDER DIFFERENCES IN PATTERNS OF RELATIONSHIP VIOLENCE IN ALBERTA
MARILYN I. KWONG and KIM BARTHOLOMEW, Simon Fraser University
DONALD G. DUTTON, University of British Columbia
Gender differences in patterns of relationship violence were investigated in a representative sample of adult men (N = 356) and women (N = 351) from the province of Alberta. Respondents reported on their receipt and perpetration of violent acts in the year prior to the survey. Men and women, respectively, reported similar one-year prevalence rates of husband-to-wife violence (12.9% and 9.6%) and wife-to-husband violence (12.3% and 12.5%). However, differential gender patterns of reporting were identified. On average, men reported that they and their female partners were equally likely to engage in violent acts and to initiate violent conflicts. In contrast, women reported lower levels of victimization than perpetration of violence, and they reported less male-only and male-initiated violence than did men. The majority of respondents in violent relationships reported a pattern of violence that was bidirectional, minor, infrequent, and not physically injurious. The discussion focuses upon the meaning of gender differences in reports of relationship violence, and the existence of distinct patterns of violence within intimate relationships.
The academic and popular literatures on relationship violence focus primarily on the battery of women by their male partners. However, a number of representative surveys in the U.S. have suggested that a large minority of both men and women commit violent acts within their intimate relationships (e.g., Morse, 1995; Straus & Gelles, 1986). These findings have raised questions about gender differences in experiences of relationship violence. Although a few Canadian surveys have assessed both male-to-female and female-to-male violence, none have included reports of perpetration and receipt of violent acts from both men and women. In addition, these surveys have provided limited information on the context and consequences of the violence reported. In the present study, we have re-analyzed data from Kennedy and Dutton's 1987 survey of Alberta, in which information on wife-to-husband violence was collected but not analyzed or reported, to investigate gender differences in experiences of relationship violence.(1) Four aspects of relationship violence were examined: rates of perpetration and receipt of violence, bidirectionality of violence, initiation of violence, and consequences of violence.
Table 1 presents a summary of the relationship violence surveys reviewed in this paper. All of these surveys employ probability sampling in which every individual in the sampling frame has a known probability of being selected. With one exception (Bland & Orn, 1986), these surveys have assessed rates of violence using variants of the Conflict Tactics Scales (CTS; Straus, 1979), the most widely used instrument for assessing relationship violence. Two violence rates can be obtained by this measure: (1) one-year prevalence rates, the proportion of participants reporting violence in the past year, and (2) prevalence rates, the proportion of participants reporting violence at any time in the past.
TABLE 1a: prevalence rates (violence experienced ever in the past)
Relationship Violence Surveys
Direction of Violence Male-to-Female Female-to-Male Survey (%) (%) Canadian Surveys: Bland & Orn (1986)(a) 14.6 22.6 Brinkerhoff & Lupri (1988)(b) 24.6 27.5 DeKeseredy & Schwartz (1998)(ade) 35.0 46.1 Grandin & Lupri (1997)(b) 18.3 25.3 Kennedy & Dutton (1989)(b) 11.2 - Srnith (1987)(be) 14.4 - Statistics Canada (1993)(ce) 15.0 - Non-Canadian Surveys: Straus & Gelles (1986)(b) Men's Reports 12.2 10.5 Women's Reports 12.1 11.9 Magdol et al. (1997)(bd) Men's Reports 22.8 34.1 Women's Reports 27.1 37.2
b: one-year prevalence rates
c: violence experienced from a current partner
d: young adult sample
e: women's reports only
Four Canadian surveys have obtained and reported rates of both men's and women's experiences of relationship violence from representative community samples (Bland & Orn, 1986; Brinkerhoff & Lupri, 1988; DeKeseredy & Schwarz, 1998; Grandin & Lupri, 1997). It is notable that the rates of male perpetrated violence obtained from these surveys are as high as or higher than those obtained in Canadian surveys that have only assessed violence against women (Smith, 1987; Statistics Canada, 1993). Furthermore, the findings of these four surveys suggest that at least as many women as men have acted violently within their intimate relationships. Unfortunately, none of these surveys obtained victimization and perpetration reports from both genders, thus preventing the comparisons required to validate men's and women's reports. To illustrate, women's perpetration reports and men's victimization reports need to be compared to validate the obtained rates of wife-to-husband violence.
The finding that similar rates of relationship violence are reported by men and women has been replicated in many representative surveys conducted outside of Canada (see Straus, 1993, for a review). In several of these studies, reports of both perpetration and receipt of violence were obtained from both men and women. Table 1 presents the reported rates for a national U.S. survey of married or cohabiting adults (NFVS; Straus & Gelles, 1986), and a survey of an unselected birth cohort of young adults in New Zealand (Magdol et al., 1997). Although it appears that young adults report higher rates of violence than older adults, men's and women's perpetration rates are still comparable.
The similar rates of perpetrated and received violence generally reported by men and women raise an important but volatile issue -- physical assault of husbands by wives. Some information on the meaning of these findings may be gained by exploring gender differences in the context and consequences of relationship violence. Context has typically been examined through assessment of the bidirectionality and initiation of violence.
In their study of Calgarian couples, Brinkerhoff and Lupri (1988) found that of 213 couples reporting violence, both partners reported perpetrating violence in 38% of couples, while in 27% only the husband reported violence and in 35% only the wife reported violence. Although DeKeseredy and Schwartz's (1998)survey of young adults included information that could address bidirectionality, the directionality data is not included in their reports of this survey. Extensive U.S. research, however, has indicated that much of the violence experienced by intimate partners is bidirectional. For example, in the 1985 NFVS, of the sub-sample of 825 respondents who disclosed violence, 49% reported that both spouses had engaged in violence, 23% reported male-only violence, and 28% reported female-only violence (Stets & Straus, 1990).
Regarding the initiation of violence, Bland and Orn (1986) found that of those individuals reporting violent relationships, 58% of men and 73% of women indicated that they had initiated violence by hitting or throwing things first, "regardless of who started the argument" (p. 131). Similarly, in the 1985 NFVS, more women (53%) than men (44%) admitted to having struck the first blow in violent confrontations (Stets & Straus, 1990). Unfortunately, little additional information about the circumstances surrounding these incidents was available to assess the context in more detail. For instance, measures of bidirectionality generally refer to the use of violence and not the level of violence. It is possible that in many bidirectionally violent relationships, men are the primary aggressors and women are responding physically in self-defense. However, findings showing female-only violence and female initiation of violence suggest that this is not always the case. Interestingly, in DeKeseredy and Schwartz's (1998) survey of young adults, 62.3% of women reporting perpetration of minor violence said it was never in self-defense and only 6.9% said it was always in self-defense. Similarly, of women reporting perpetration of severe violence, 56.5% said it was never in self-defense and 8.5% said that it was always in self-defense.
Figure 1. Gender Comparisons for Reports of Relationship Violence
While gender differences in the context of relationship violence are unclear, U.S. data on the consequences of relationship violence have shown that women are more likely to suffer physical injury and seek medical attention than men. The 1985 NFVS found that 3% of female victims, but only 0.4% of male victims, reported that they required medical care for injuries sustained from partner violence (Stets & Straus, 1990). A survey of young adults found that women in violent relationships feared physical injury significantly more than men (Morse, 1995). Thus, it seems that while both men and women engage in frequent minor assault, women are more likely to sustain and fear physical injury.
The present study used the Alberta data collected, but not reported, by Kennedy and Dutton (1989) to explore gender differences in experiences of relationship violence. As far as we are aware, this is the only data available from a representative Canadian survey to assess gender differences in the rates, context, and consequences of violence in intimate relationships. In contrast to Kennedy and Dutton's (1989) procedure of combining reports by both genders to derive wife assault rates, we have presented separately male and female reports of the perpetration and receipt of violent acts. In addition to permitting gender comparisons, separate reports enabled us to check for agreement between men and women on rates of received and perpetrated violence. We also compared men's and women's reports of bidirectionality of violence and initiation of violence. In order to better understand the context of violence, descriptive analyses on gender differences in the types and frequencies of specific acts of relationship violence were conducted. Finally, we compared male and female reports on the consequences of violence.
As part of a larger population research project, 707 adult residents (356 men and 351 women) of the province of Alberta completed a relationship violence survey in1987 (Kennedy & Dutton, 1989). To be eligible for the violence component of the study, participants had to have been living in a marital or "marriage-like" relationship in the year prior to the survey. See Kennedy and Dutton (1989) for information on the demographic characteristics of the sample and the sampling and weighting procedures.
Violence was assessed with a short form of the Conflict Tactics Scales (CTS; Straus, 1979). This measure was comprised of nine violence items: threw something; pushed, grabbed, or shoved; slapped; kicked, bit, or hit with a fist; hit or tried to hit with something; beat up; choked; threatened with a knife or gun; used a knife or gun. Kennedy and Dutton (1989) omitted the choking item in their analyses to remain consistent with the NFVS surveys (e.g., Straus & Gelles, 1986). However, we have included the item in our analyses. Respondents were asked how many times each of these acts had occurred in the last 12 months. First, they reported on their own behaviour, then on their partner's behaviour. Raw frequency counts were recorded.
Scoring of One-year Prevalence Rates. Three indices of violence were derived from the CTS items. An overall violence index was derived from responses to all nine items. Minor and severe violence indices were derived from the first three and last six items respectively. For example, the one-year prevalence rate of overall violence indicates the percentage of respondents who reported at least one occurrence of any of the items included in the overall violence scale during the year prior to the survey, and the one-year prevalence rate of severe violence refers to the percentage of respondents who endorsed at least one item on the severe violence index. Thus, rates indicate the existence of violence only. They do not give any information about the frequency of violence reported by respondents.
To assess initiation of violence, respondents were asked who had started the physical conflict in the most serious physical fight they had had with their partner in the past year. Response choices were "you"' or "spouse/partner," although responses of "both" or "neither" were recorded if volunteered.
To assess the consequences of violence, respondents reporting receipt of violence were asked whether they had ever (1) sustained serious injuries, (2) had to go to a hospital's emergency ward for treatment, and (3) had to take time off from work because of a physical fight with a partner. Response choices were "yes" or no."
One-year Prevalence Rates for Specific Acts of Violence by Gender of Respondent
Husband-to-Wife Wife-to-Husband Male
Type of Violence (%) (%) (%) (%) A. Minor Violence Acts Threw something 5.1 3.4 6.5 4.8 Pushed/grabbed/shoved 10.4 7.7 9.6 8.8 Slapped 1.4(c) 2.0 4.2 4.3(d) B. Severe Violence Acts Kicked/bit/bit 0.0(c) 1.7 3.1 1.4(d) Hit/tried to hit with something 2.0 1.1 2.5 3.4 Beatup 0.0 0.6 0.0 0.0 Choked 0.3 0.6 0.8 0.3 Threatened with knife/gun 0.0 0.3 0.3 0.9 Used knife or gun 0.0 0.0 0.0 0.3
a: N = 356;
b: N = 351.
Note: Across rows, proportions with differing subscripts are significantly different at p < .05.
Four sets of analyses are presented: (1) one-year prevalence rates of unidirectional violence (husband-to-wife and wife-to-husband), (2) bidirectionality of violence, (3) initiation of violence, and (4) consequences of violence. Male and female reports are presented separately to permit gender comparisons. Figure 1 displays the various gender comparisons made for rates of unidirectional violence. Comparisons were also made between men's and women's reports of the bidirectionality, initiation, and consequences of violence.
One-year Prevalence Rates
Gender agreement on one-year prevalence rates. Figure 2 presents one-year prevalence rates for the three indices of husband-to-wife violence. The proportions of male respondents who reported overall, minor, and severe husband-to-wife violence were 12.9%, 12.1%, and 2.2%. The proportions of female respondents who reported overall, minor, and severe partner violence were 9.6%, 9.6%, and 2.8%. Although men reported higher rates of husband-to-wife violence than women, tests comparing the independent proportions were not significant. Al in Figure 1 illustrates the comparison made to test gender agreement on husband-to-wife violence.
Figure 2. One-year Prevalence Rates for Relationship Violence by Gender of Respondent
Figure 2 also shows one-year prevalence rates for the three indices of wife-to-husband violence. The proportions of female respondents who reported overall, minor, and severe wife-to-husband violence were 12.5%, 1l.9%, and 4.5%. The proportions of male respondents who reported overall, minor and severe partner violence were 12.3%, 12.1%, and 4.8%. A2 in Figure 1 illustrates the comparison made to test gender agreement on wife-to-husband violence.
In summary, men and women agreed on one-year prevalence rates of both husband-to-wife and wife-to-husband violence. Moreover, the obtained rates were similar to those found in the 1985 NFVS (Straus & Gelles, 1986).
Gender comparisons of one-year prevalence rates. In contrast to the previous analyses of gender agreement, the next set of gender comparisons examines whether a higher proportion of one gender experiences violence than the other. First, within-gender reports were compared to see whether each gender reported perpetrating and receiving differing levels of violence. See Figure 1 for an illustration of these two comparisons: B1 indicates within-male reports, and B2 within-female reports. As presented in Figure 2, men reported perpetrating and receiving comparable levels of overall and minor violence. However, men reported receiving significantly more severe violence than they perpetrated (4.8% versus 2.2%, z = 2.32, p < .05).
In contrast, female reports in Figure 2 show that women reported receiving lower levels of violence than they perpetrated for all three indices. However, only the difference between rates of overall violence perpetrated and received was significant (z = 2.2, p < .05).
Next, we made two across-gender comparisons: one for reports of violence received (Figure 1, Comparison C), and another for reports of violence inflicted (Figure 1, Comparison D). Comparison C indicated that men reported receiving more violence than women reported receiving on all three indices. However, tests of independent proportions were not significant for any of these differences. There was no clear pattern for Comparison D between men's and women's reports of violence inflicted, and there were no significant differences.
Gender comparisons of specific acts of violence. Table 2 presents one-year prevalence rates for specific acts of violence. According to both men's and women's reports, minor acts of violence were generally more common than acts of severe violence. Furthermore, men's and women's rates for each specific act of violence were similar except for two items, "slapped" and "kicked/bit/hit", where significantly more women than men reported perpetrating these acts (z = 2.3, p < .05 and z = 2.4, p < .05, respectively).
In summary, the results of within-gender comparisons were inconsistent with across-gender comparisons. Within-gender comparisons showed that men reported receiving more severe violence than they inflicted, and women reported inflicting more overall violence than they received. However, no significant differences in violence experienced by men and women were found using across-gender reports. This inconsistency is probably due to the greater statistical power of the within-gender test (i.e., tests of differences between dependent samples are more powerful than tests of independent samples). As Figure 2 shows, across-gender reports for wife-to-husband violence (male report: 12.3%; female report: 12.5%) are consistent, and male-reports for violence perpetrated and received (12.9% and 12.3%) are consistent. However, women's reports of the violence they received (9.6%) are lower than their reports of violence inflicted (12.5%) and lower than men's reports of violence inflicted (12.9%).
Descriptive analyses were conducted to examine the proportions of respondents who reported experiencing one of three levels of violence: low (1 to 4 incidents), moderate (5 to 20 incidents), and high (more than 20 incidents). These cutoffs were based on an examination of the distributions of violence frequency and a consideration of the likely implications of different frequencies of violence. Only respondents who reported experiencing at least one incident of violence in the year prior to the survey were included in these analyses. As shown in Figure 3, the majority of perpetrators (male report: 85%; female report: 69%) and victims (male report: 77%; female report: 79%) of violence reported relatively low levels of violence. Few perpetrators (male report: 4%; female report: 9%) and victims (male report: 5%; female report: 12%) reported high levels of violence. There were no significant gender differences in violence levels reported by either perpetrators or victims.
Figure 3. Gender Comparisons of Levels of Violence
Bidirectionality of Violence
The data of respondents who experienced any violence were categorized by who had engaged in physical conflict: both partners, male only, or female only. Of those women who reported any violence in the past 12 months (received and/or inflicted), 52% reported violence from both partners, 35% reported female-only violence, and 13% reported male-only violence. Of the men who reported any violence in the past 12 months (received and/or inflicted), 62% reported that both partners were violent, 18% reported female-only violence, and 20% reported male-only violence. Thus, according to more than one-half of men and women in relationships with any violence, both partners had engaged in at least one violent act. Furthermore, women were significantly more likely than men to report female-only violence (z = 2.0, p < .05).
Bidirectionality was also assessed by comparing reports of perpetration and receipt of violence within gender to test for the likelihood of bi-directional violence. Sixty percent of women who reported perpetrating violence also reported receiving violence; conversely, 79% of women who reported receiving violence also reported perpetrating violence (c2(1, n = 354) = 150.8, p < .01). Similarly, 76% of men who reported perpetrating violence also reported receiving violence, and 77% of men who reported receiving violence also reported perpetrating violence (c2(1, n = 356) = 189.9, p < .01). Parallel analyses also indicated strong dependencies in reports of perpetration and receipt of minor and severe violence, for both women and men.
Initiation of Violence
Male and female respondents who had reported any violence within their relationship were asked who had initiated the physical conflict during the most serious incident that occurred in the prior year. Of men reporting violence, 49% identified themselves as the initiator, 35% identified their partner, and 14% identified both partners. Of women reporting violence, 67% identified themselves as the initiator, 27% identified their partner, and 6% reported that both had initiated the conflict. Although more women than men appeared to attribute the initiation of violence to themselves (a difference of 18%), men's and women's proportions for reported self-initiation did not differ significantly. Nor was there a significant difference between men's and women's reports of partner initiation. However, significant differences were found when checking for gender agreement in reports for female and male initiation. Women were significantly more likely than men to report female initiation (z = 3.2, p < .01), and men were significantly more likely than women to report male initiation (z = 2.3, p < .05).
Consequences of Violence
Of the 19 women who were asked about the consequences of a physical fight with their partner,(2) only three reported that they had experienced any of the three consequences assessed in the survey (serious injury, need for medical attention, and time off work). Two women reported two consequences and a third woman endorsed all three consequences. No men reported experiencing any of these consequences. The low endorsement rate on the consequence items does not necessarily mean that respondents who reported violence experienced no injurious consequences. The consequence items used in this survey assessed only relatively severe physical consequences and did not include psychological consequences of violence.
A more in-depth analysis indicated that women who reported suffering consequences from violence also reported experiencing the highest frequencies of abuse, both as perpetrators and victims. These frequencies refer to the total number of violent behaviours reported for the 12 months prior to the survey. For example, if a woman reported being slapped 10 times and kicked 15 times, she would receive a frequency score of 25. Two of these women reported receiving much more violence than they inflicted: One woman reported that her partner had perpetrated 49 violent acts against her while she had engaged in 23 such acts, and the other woman reported that she had been the victim of 218 violent acts while she had engaged in 72 such acts. A third woman reported inflicting many more acts of violence than she received. She reported perpetrating 297 violent acts against her husband and receiving 22 violent acts. All three women reported that it was their partner who had initiated the physical conflict during the most serious incident in the year before the survey. The pattern of violence described by the two women who reported receiving more violence than they perpetrated could represent women attempting to defend themselves from consistently abusive partners. Although such relationships would be categorized as bidirectionally violent, the frequency of received violence between partners is dearly different. This illustrates why it is critical not to interpret reports of the existence of bidirectional violence as indicating equivalence in the severity or nature of violence experienced.
One-year Prevalence Rates
Based on Kennedy and Dutton's (1989) representative survey of Alberta, the proportions of men and women who reported perpetrating and receiving at least one act of violence within their intimate relationships in a 12-month period were roughly comparable. Specifically, rates of overall husband-to-wife violence were 12.9% and 9.6%, and rates of overall wife-to-husband violence were 12.3% and 12.5%, according to men's and women's respective reports. There were no significant gender differences in reported rates for each category of violence, indicating gender agreement. Furthermore, with only two exceptions, similar proportions of men and women tended to report engaging in specific acts of violence that were usually minor and infrequent. The comparable one-year prevalence rates of male and female violence are consistent with the findings of a number of non-Canadian studies (e.g., Straus & Gelles, 1986). In addition, the obtained rate of male violence is generally consistent with prior Canadian surveys that only included women's reports of male violence (e.g., Smith, 1985, 1987; Statistics Canada, 1993). Hence, the consistency of rates across men's and women's reports and across different studies supports the validity of the current findings.
Context and Consequences
Consistent with prior research (e.g., Morse, 1995; Stets & Straus, 1990), measures of the context of violence revealed that much of the violence experienced by these men and women was bidirectional. Sixty-two percent of men and 52% of women who reported violence indicated that it was perpetrated by both partners. As discussed previously, evidence of bidirectional violence does not inform us of the relative severity of male-to-female and female-to-male violence. However, some insight into this issue can be obtained by comparing the frequencies of perpetrated and received violent acts for individual indicating bidirectional violence. It could be, for instance, that women reporting bidirectional violence report higher levels of receipt than perpetration, suggesting that their perpetration may be self-defensive. We defined asymmetric violence as those cases in which a participant reported that one partner had inflicted at least five more acts of violence than the other partner in the past year. Based on this criterion, 15 of the 62 respondents reporting bidirectional violence were classified as experiencing asymmetric violence. Of the 7 men classified as asymmetric, 6 reported receiving more violence than they perpetrated and one reported the opposite pattern. Of the 8 women classified as asymmetric, 3 reported receiving more violence than they perpetrated and 5 reported the opposite pattern. These small numbers preclude any definitive interpretation. However, these data indicate that only a minority of respondents reported clearly asymmetric patterns of violence in their relationships. Moreover, we found no evidence that male-to-female violence was more frequent than female-to-male violence in bidirectional cases.
Surprisingly, both men and women tended to attribute the initiation of violence to themselves. Specifically, 49% of men and 67% of women reported that they had started the physical conflict. Since participants were given only forced choice response options, however, it is not clear how reports of initiation should be interpreted. Participants were not asked about their reasons for initiating physical violence or about the role of psychological aggression during the incident.
The overwhelming majority of respondents who experienced violence did not endorse any consequence items. Just 16% of women who reported receipt of violence endorsed one or more of the consequences assessed, and no men reported experiencing these consequences from their receipt of violence. This pattern of findings is similar to that from the 1985 NFVS: Of the participants reporting receipt of violence, 3% of the women, and only 0.4% of the men, required medical care for injuries sustained as a result of domestic violence (Stets & Straus, 1990).
Gender Patterns in Violence Reports
Within-gender analyses of violence reports showed that men generally reported gender equality in the perpetration, receipt, initiation, and bidirectionality of abuse. In contrast, women tended to report that they were more likely to perpetrate violent acts than their male partners. Furthermore, a smaller proportion of women reported male-only violence (13%) compared to female-only violence (35%), and fewer women reported male initiation of violence (26%) than female initiation of violence (67%). At the very least, this data suggests that not all of women's violence within intimate relationships can be interpreted as self-defensive (cf., Straus, 1993).
An across-gender comparison of men's victimization reports and women's perpetration reports showed gender agreement for rates of wife-to-husband violence. Other across-gender comparisons, however, revealed some discrepancies. As shown in Figure 2, all reports of overall violence were consistent except for women's reports of violence they received. Moreover, women reported significantly more female-only violence and female initiation of violence than men. in contrast, men reported significantly more male initiation of violence than women.
This pattern of findings is contrary to common expectations that it is men who are motivated to understate their perpetration of violence. These observed discrepancies may be due to women underreporting the violence that they receive, perhaps in an attempt to protect their partners or to minimize the problem. Alternately, men may be overreporting the violence they perpetrate, perhaps to avoid appearing to be a victim or to compensate for the guilt they feel from being abusive. Of course, both genders' reports of violence may be biased: Men may be overreporting to some degree, and women may be underreporting to some degree. For a more thorough discussion of potential sources of reporting biases, see Moffitt et al. (1997).
Selection biases may also account for the observed discrepancies in men's and women's reports of violence. As has often been argued, abused women and violent men may be reticent to participate in relationship violence surveys. If abused women are even less likely to be included than violent men, women's reports of the receipt of male violence will be lower than men's reports of their perpetration of violence. However, it seems equally plausible that male victims of violence (and perhaps, violent women as well) will be underrepresented in these surveys. In addition, the sampling method may have systematically excluded some individuals at high risk for being assaultive or being the victim of assault. In particular, male batterers who are incarcerated and female victims in transition homes were not included in the sampling frame, thereby potentially resulting in an underestimate of the rate of severe husband-to-wife violence.
Patterns of Relationship Violence
Johnson (1995) has proposed that the phenomenon of "domestic violence" can be broken down into two distinct patterns that he labels patriarchal terrorism and common couple violence. Patriarchal terrorism describes violence in which women are highly victimized by husbands who seek control over them. Common couple violence represents occasional violence that is often enacted and initiated equally by men and women during conflicts that get out of hand. Johnson argues that studies of women seeking aid for domestic violence, such as women in shelters, generally include only women experiencing the extreme form of domestic violence. It is therefore inappropriate to generalize from these studies to the entire population of violent couples. On the other hand, representative surveys reach individuals experiencing common couple violence, but underrepresent those experiencing extreme forms of domestic violence, because both highly violent and highly victimized individuals are unlikely to participate in this type of survey. Similarly, Straus (1990, 1993) has discussed the "clinical fallacy" of generalizing from studies of women identified as battered to the general population and the "representative sample fallacy" of generalizing from representative samples to specific subsamples such as severely battered women.
There is some support for these distinctions in the present study. Only 3 of the 52 women who reported receiving any violence in the year prior to the survey fit the batterer/victim pattern of clearly asymmetrical violence. The majority of violence reported by respondents was equally perpetrated by men and by women, relatively minor and infrequent, and did not result in injury. However, a significant proportion of the subsample reporting violence described patterns of violence that did not fit either the patriarchal terrorism or common couple violence pattern A minority of men and women reporting violence indicated receiving (18% and 9%, respectively) and perpetrating (11% and 22%, respectively) moderate levels of violence in their relationships, and in many cases it was difficult to determine whether the reported violence was mutual or not. Although the typology outlined by Johnson (1995) is helpful, we suggest that researchers and clinicians be open to a range of abuse patterns that differ in frequency, severity, and direction.
Limitations and Future Research
Unfortunately, there is little information in the present study on potential gender differences in the meaning of violence or the interpersonal dynamics associated with violence. The measures of violence available -- one-year prevalence rates, bidirectionality of violence, initiation of violence, and consequences of violence -- are inherently impoverished. In particular, the original CTS has a number of limitations (e.g., Dobash, Dobash, Wilson, & Daly, 1992). Assessments of the context and consequences of relationship violence could be enhanced by including measures of psychological abuse and psychological consequences. Research has shown that psychological abuse may be as harmful as physical violence (Herbert, Silver, & Ellard, 1991) and that women in mutually violent relationships report more negative psychological impact from their receipt of violence than do their husbands (Vivian & Langhinrichsen-Rohling, 1994). Finally, the present sample only assessed relationship violence in the past year.
Another limitation of this survey is that it is not able to address gender differences in the psychosocial correlates of relationship violence. Surprisingly little research has examined this question, though there are some indications that the developmental pathways leading to relationship aggression may differ for men and women. For example, Capaldi and Crosby (1997) found that antisocial behaviour was predictive of relationship aggression for young men, whereas depressive symptoms and low self-esteem were predictive for young women. In contrast, Magdol et al. (1997) found that various psychosocial variables, including antisocial behaviour and depression, were predictive of young men's violence, but not young women's.
An additional limitation of the current survey is that only men and women in marriage-like relationships were sampled. Excluding people who are single, divorced, or separated may have deflated rates of reported violence since higher rates of relationship violence have been found in samples of dating couples (DeKeseredy & Schwartz, 1998) and separated or divorced women (Schwartz, 1989). Furthermore, the selection criteria for the current study (living in a "marriage like relationship" in the past year) was oriented towards heterosexual men and women, and this doubtless led to a systematic underrepresentation of men and women in same-sex relationships. Because of this selection criteria and the fact that 91% of respondents indicated that they were currently or previously married, we have interpreted the data in terms of heterosexual relationships. However, it is possible that, in some cases, people are reporting the perpetration and receipt of violence with same-sex partners. For future surveys, it is important to include all adults in the sampling frame, to word all survey questions and selection criteria in more inclusive terms, and to explicitly assess sexual orientation.
While these improvements could address some of the limitations of survey research, telephone questionnaires are best suited for collecting data that is standardized and objective. Context and meaning, on the other hand, are full of nuances and subjectivity. Therefore, more in-depth interviewing techniques may be better suited for investigating the context of violence. Future research could benefit from the strengths of both methods by employing representative surveys to obtain estimates of rates and then administering in-depth interviews to a subsample of respondents to explore the more complex relationships between abuse and its contexts and consequences. In-depth interviews are particularly important in understanding how the meaning and consequences of relationship violence may differ for men and women.
This is the first Canadian study to investigate gender differences in the rates, bidirectionality, initiation, and consequences of relationship violence in a representative sample. Consistent with research outside of Canada, men and women reported similar rates of violence perpetration and victimization. And, while more comprehensive study is needed, it appears that a substantial proportion of women's violence cannot be explained as acts of self-defense. Both genders reported that women do initiate violence and are sometimes the sole perpetrators of aggression in relationships. Also consistent with prior research, the violence reported by respondents generally differed from the prototypical batterer/victim pattern. The majority of respondents in violent relationships reported a pattern of violence that was bidirectional, minor, infrequent, and not physically injurious.
While the importance of eliminating violence against women is obvious, the need to stop women's violence against men may be less evident. Our society seems to harbour an implicit acceptance of women's violence as relatively harmless, even amusing. For example, the popular media frequently shows women hitting men with little consequence. Straus (1993) has outlined four reasons in support of the elimination of women's violence: 1) spousal assault is morally wrong, regardless of gender; 2) the acceptance of female violence may perpetuate traditional norms tolerating violence between intimate partners; 3) there is evidence that women's violence may increase the probability of spousal conflict escalating into severe wife battery; and 4) all forms of spousal violence model violence to children and may be predictive of children subsequently being perpetrators or victims of relationship violence themselves. We would add that although a low level of violence between men and women may not be physically injurious, it is associated with high marital and individual distress (e.g., Vivian & Langhinrichsen-Rohling, 1994). Furthermore, the failure to acknowledge the possibility of women's violence, in the face of sound research evidence, jeopardizes the credibility of all theory and research directed toward ending violence against women. It also does an injustice to men who are victims of female violence and to women who need help in learning more constructive strategies to deal with the inevitable conflicts and frustrations that arise in intimate relationships.
As Straus (1990, 1993) has emphasized, findings indicating that many men and women are involved in relationships with low-level, often mutual, violence do not negate the problem of severe relationship violence in which men are likely the primary perpetrators and women the primary victims. Nor does the small proportion of survey respondents reporting an extreme batterer/victim pattern negate the usefulness of survey methodology for investigating relationship violence. The finding that a sizable minority of Albertan men and women reported at least one incident of violence in a one-year period shows that a considerable amount of physical aggression does occur in intimate relationships. And, because low level violence is associated with marital and individual distress, some of these men and women may seek some form of therapy. For example, in a sample of maritally distressed couples seeking marital treatment, Cascardi, Langhinrichsen, and Vivian (1992) found a prevalence rate of 71% for marital aggression. Thus, marital therapists should be prepared to deal with different types of relationship violence.
Unfortunately, we do not yet have sufficient information to fully describe the range of relationship violence that Canadians experience, nor the nature or meaning of that violence. However, it is becoming clear that the mutual/low and batterer/victim patterns of relationship violence may be largely non-overlapping. These distinct patterns, and other possible patterns yet to be identified, may represent different problems that require different therapeutic approaches. Couples who are experiencing low levels of relationship violence may seek therapy, but not specifically for the physical conflict. In fact, spouses rarely report physical aggression spontaneously when asked about the marital problems for which they are seeking help (O'Leary, Vivan, & Malone, 1992). Therefore, efforts directed at improving communication and conflict resolution strategies may be more beneficial than focusing solely on the issue of violence. However, it is important to keep in mind that, within the criminal justice system, any of the physical acts endorsed by these respondents would constitute assault. Further research is required to enhance our understanding of all forms of violence within intimate relationships and to guide our efforts in appropriate preventative and therapeutic approaches.
The authors would like to thank Leslie Kennedy for generously making this data available to us and Stephen Hart for his feedback during the preparation of this paper.
Bland, R., & Orn, H. (1986). Family violence and psychiatric disorder. Canadian Journal of Psychiatry, 1, 129-137.
Brinkerhoff, M.B., & Lupri, E. (1988). Interspousal violence. Canadian Journal of Sociology, 13, 407-434.
Capaldi, D.M., & Crosby, L. (1997). Observed and reported psychological and physical aggression in young, at-risk couples. Social Development, 6, 184-206.
Cascardi, M., Langhinrichsen, J., & Vivian, D. (1992). Marital aggression: Impact, injury and health correlates for husbands and wives. Archives of Internal Medicine, 152, 1178-1184.
Cascardi, M., & Vivian, D. (1995). Context for specific episodes of marital violence: Gender and severity of violence differences. Journal of Family Violence, 10, 265-293.
DeKeseredy, W.S., & Schwartz, M.D. (1998). Woman Abuse on Campus: Results from the Canadian National Survey. Thousand Oaks, CA: Sage Publications.
Dobash, R.P., Dobash R.E, Wilson, M., & Daly, M. (1992). The myth of sexual symmetry in marital violence. Social Problems, 39, 71-91.
Grandin, E., & Lupri, E. (1997). Intimate violence in Canada and the United States: A cross-national comparison. Journal of Family Violence, 12, 417 - 443.
Herbert, T.B., Silver, R.C., & Ellard, J.H. (1991). Coping with an abusive relationship: How and why do women stay. Journal of Marriage and the Family, 53, 311-325.
Hotaling, G.T., & Sugarman, D.B. (1986). An analysis of risk markers in husband to wife violence: The current state of knowledge. Violence and Victims, 1, 101-124.
Johnson, M.P. (1995). Patriarchal terrorism and common couple violence: Two forms of violence against women. Journal of Marriage and the Family, 57, 283-294.
Kennedy, L.W., & Dutton, D.G. (1989). The incidence of wife assault in Alberta. Canadian Journal of Behavioural Science, 21, 40-53.
Magdol, L., Moffitt, T.E., Caspi, A., Newman, D.L., Fagan, J., & Silva, P.A., (1997). Gender differences in partner violence in a birth cohort of 21-year-olds: Bridging the gap between clinical and epidemiological approaches. Journal of Consulting and Clinical Psychology, 65, 68-78.
Moffitt, T.E., Caspi, A., Krueger, R.F., Magdol, L., Margolin, G., Silva, P.A., & Sydney, R. (1997). Do partners agree about abuse in their relationship?: A psychometric evaluation of interpartner agreement. Psychological Assessment, 9, 47-56.
Morse, B. J. (1995). Beyond the Conflict Tactics Scale: Assessing gender differences in partner violence. Violence and Victims, 10, 251-272.
O'Leary, K.D., Vivian, D., & Malone, J. (1992). Assessment of physical aggression in marriage: The need for a multimodal method. Behavioral Assessment, 14, 5-14.
Schwartz, M.D. (1989). Asking the right questions: Battered wives are not all passive. Sociological Viewpoints, 5, 46-61.
Smith, M.D. (1985). Woman abuse: The case for surveys by telephone. LaMarsh Research Programme on Violence and Conflict Resolution, Report No. 12. York University, North York, Ontario, Canada.
Smith, M.D. (1987). The incidence and prevalence of woman abuse in Toronto. Violence and Victims, 2, 173-187.
Smith, M.D. (1990). Sociodemographic risk factors in wife abuse: Results from a survey of Toronto women. Canadian Journal of Sociology, 15, 39-58.
Statistics Canada (1993). The Violence Against Women Survey. The Daily, November 18, 1993. Ottawa, ON. Ministry of Industry, Science and Technology.
Stets, J.E., & Straus, M.A. (1990). Gender differences in reporting marital violence and its medical and psychological consequences. In M.A. Straus & R.J. Gelles (Eds.), Physical Violence in American Families: Risk Factors and Adaptations to Violence in 8,145 Families (pp. 227-244). New Brunswick, NJ: Transaction Publishing.
Straus, M.A. (1979). Measuring intrafamily conflict and violence: The Conflict Tactics (CT) Scales. Journal of Marriage and The Family, 41, 75-88.
Straus, M.A. (1990). Injury and frequency of assault and the "Representative Sample Fallacy" in measuring wife beating and child abuse. In M.A. Straus & R.J. Gelles (Eds.), Physical Violence In American Families: Risk Factors and Adaptations To Violence in 8,145 Families (pp. 75-91). New Brunswick, NJ: Transaction Publishing.
Straus, M.A. (1993). Physical assaults by wives: A major social problem. In R.J. Gelles & D. Loseke (Eds.), Current Controversies in Family Violence (pp. 67-87). Newbury Park, CA: Sage.
Straus, M.A., & Gelles, R.J. (1986). Society change and change in family violence from 1975 to 1985 as revealed by two national surveys. Journal of Marriage and the Family, 48, 465-479.
Vivian, D., & Langhinrichsen-Rohling, J. (1994). Are bi-directionally violent couples mutually victimized?: A gender-sensitive comparison. Violence and Victims, 9, 107-124.
Received October 14, 1997
Revised September 9, 1998
Accepted September 22, 1998
1. The original researchers did not look at gender differences because they were primarily interested in male-female violence at the time.
2. Unfortunately, only 19 of the 34 women reporting the receipt of violence were asked this question. This was an error in the administration of the survey. There were no differences in violence reports between those women who were and were not asked the consequences question.